FR2825179A1 - VOICE RECORDING METHOD AND DATA CARRIER - Google Patents

Abstract

A method of voice introduction of a word with a microphone and comparison of the voice signal with the recorded phonetic data. The user first introduces a selection criterion that limits the phonetic data to be examined. From this comparison, the name entered is determined. The invention makes it possible to increase the reliability of the vocal introduction and to accelerate this introduction.

Description

information about the windshield of the vehicle.

  i. State of the Art The invention relates to a method for introducing a voice into

  word that the voice signal is captured by a microphone.

  Navigation devices are known for vehicles intended to allow guidance in a road network, according to which the destination can be vocally introduced. In this system, the letters are pronounced one by one, and with the help of them the identification identifies itself, but

  however, the wording of the letters is long.

  According to the document DE 197 42 054 A1, there is known a system of entry for localities and / or street names in a navigation system according to which the entry is done by whole words. These words must, however, be recognized by a voice recognition device and then search in an alphabetical list for the name of place and / or street. It is necessary for this that the whole word is perfectly

  recognized by the voice recognition device. The recognition

  hold can be done with error and requires a very important data state

possible vocal models.

  Advantages of the invention The present invention relates to a method of the type defined above, characterized in that phonetic data is recorded with at least one complementary piece of information, the user introduces a

  criterion of selection, the criterion of selection introduced is compared with the infor-

  In addition, the phonetic data are compared

  to the voice signal for which the selection criterion corresponds to: the additional information recorded and from the comparison,

defines the word introduced.

  The method according to the invention offers the advantage of recording

  voice models for possible complete word entries, and

  thus mimic the search among these vocal models by indicating previously a selection criterion. Thus directed research can be done

  according to the vocal patterns recorded in the form of phonetic

for complete words.

  These searches are preceded by the entry of a selection criterion.

  This is because the voice models concerned by the selection criterion will be examined in more detail. Thus, not only is the risk of a mistake of recognition avoided, because the possible words ringing in a similar way but which do not meet the selection criterion will be excluded from the outset; in addition, it is advantageous that the reduction in the importance of the search also reduces the transfer of data between the computing unit and the memory unit to the data alone.

  records for speech recognition.

  According to another characteristic, it is particularly advantageous that by the phonetic indication of the data / places and / or names of registered names, there is a direct association between the phonetic data and the indications of places and / or indications.

names.

  It is furthermore advantageous that the selection criterion is also entered by a voice input, so that one can have a complete voice input, notably during the movement with the vehicle,

  so that the driver can concentrate on maneuvers.

  It is also advantageous to record as additional information, membership of a thematic area, a region and / or a place. By the selection criteria evoked, the user

  can rapidly and efficiently limit research, especially in the use of

  reading the terms he knows. For a thematic area, the user can for example limit his entry to hotels or restaurants. This avoids any confusion with designations of sounding places in a similar way. The indication of belonging to a region or city or

  a street can greatly limit the search and thus increases the probability of

voice recognition.

  For example, as complementary information, one

  at least a part of a postal code of a place.

  It is furthermore advantageous to record the frequency of the selection until now of a word or complementary information. It is thus possible to first look for indications that have been given very frequently in the past, since these indications also have a higher probability of use for the future. If we look for everything

  So first these indications, we shorten the search time.

  It is furthermore advantageous to record on a phonetic data medium the data of the words to be introduced by introduction.

  voice in the navigation device. With the help of such a support

  the conditions at the reconnaissance unit can be reduced.

  in the navigation device, as it suffices to perform a comparison

  reason between the recorded phonetic data and the voice signal introduced

  Duit. It is also advantageous to record phonetic data directly with place and / or name naming to facilitate both data updating and access.

to these.

  It is also advantageous for the data carrier to be rewritable, because it can thus be supplemented by the most

  such as a new street name.

  It is also advantageous if the phonetic data recorded on the data carrier is acoustically transmitted, since the phonetic data can be used for a control emission of what has been inputted or for other acoustic emissions.

  especially for the issuing of indications destince to the driver.

  Drawings The present invention will be described in more detail with the aid of an exemplary embodiment shown in the accompanying drawings, in which: FIG. 1 shows a navigation device of a vehicle with a data carrier according to FIG. invention for the implementation of the speech input method, - Figure 2 shows a method according to the invention of voice input, - Figure 3 shows an embodiment of registration of donation.

  on a data carrier according to the invention.

  Description of the exemplary embodiment

  The method according to the invention of voice input of a word can be used in different devices with voice input such as computer equipment. In particular, it is advantageous to apply it to the vehicle, as these may have different users and, above all, a learning time which is part of the voice input system is required and this offers difficulties of acceptance. . Moreover, when introducing texts into voice input systems associated with computers, there is usually a context, so that a sentence of which a word has not been understood can be completed at least partly correctly by the voice input system. When introducing a destination for which the user can sometimes only express a single word to know the destination itself, the comparison with a context is not

  possible, so that we often encounter errors.

  The method according to the invention described hereinafter in its application a navigation system for driving a vehicle in a vehicle.

  road bucket by issuing directions.

  FIG. 1 shows a navigation device 1 comprising a loudspeaker 2, a display unit 3, a microphone 4 and a control unit 5 and a keyboard 6. The navigation device 1 comprises a calculation unit 7 connected by a first radio interface 18 to perform the location, that is to say, to determine the position of the vehicle preferably by a satellite signal system including GPS o system (global positioning system). The control unit 5 or the mi cro 4 allow to introduce a destination; the computing unit 7 then determines a path between the starting point and the destination, i.e. between the position detected by the first radio interface 18 and the introduced destination, using the digital map recorded on a data carrier 1 placed in the data carrier reader 9. The navigation device 1 is furthermore preferably connected to vehicle sensors 10, for example a path measurement sensor, a rotation sensor and / or a compass. for the location to the extent that it is not possible to establish contact with a satellite by the first radio interface 18. The o path calculation can be done in addition to or only in a service center 11 to which the device 1 can be connected by a mobile radio link 12 using a second radio interface 13. The display unit 3 is preferably made as a liquid crystal display

  and is installed in the center console of the vehicle.

  The microphone 4 is preferably in front of the driver, for example at the control panel, so that while driving, the driver who is normally standing can make a vocal introduction. The loudspeaker 2 is preferably a loudspeaker of the car radio installation of the vehicle. According to a preferred embodiment, the car radio installation can also be integrated in the navigation device 1. The mobile radio link 12 is in particular a mobile telophone link. The data carrier 8 is preferably a magnetic and / or optical data carrier which can be extracted from the data carrier reader 9. The production of the data carrier 8 3s as a mass memory is also possible for example under the form of a hard disk. The data support reader is designed as a reading and / or recording device, which makes it possible to record the current information transmitted by the service center 11 on the data carrier 8 with the donation medium reader.

born 9.

  For the transmission of the route indications to the driver, the calculation unit 7 uses the phonetic data recorded on the data carrier 8 and without having to calculate a voice emission, it transmits the phonetic data recorded by the loudspeaker. 2 to the driver. At the beginning or during a movement, the user indicates a destination by the keyboard 6 of the control unit 5 or by the microphone o 4 as voice input. For the control, the entered destination is displayed for the user on the display unit 3, and / or the speaker 2, emits an acoustic confirmation. During the movement, the display unit 3 and / or the speaker 2 give path indications according to the current position of the vehicle. It is advisable to do voice input when the driver is already on the road. Moreover in the vehicle generally for reasons of limited space, it is not said computer keyboard, so that the introduction using the keyboard 6 is usually complicated. The vocal introduction according to the invention is such that the computing unit 7 compares an audio signal introduced by the microphone 4 and phonetic data recorded on the data carrier 8, that is to say for example a voice model or pronunciation examples in the form of recorded voice signals. An example of a re ali satio n of the process according to the invention will be explained using the fi

gure 2.

  2s The voice introduction method begins with a call step 20 during which the user actuates for example one of the keys 6 of the control unit 5 to signal to the computing unit 7 that will perform a voice input. In the next first selection step 21, the user at the request of the calculation unit 7 made by the loudspeaker 2 or by the display unit 3, introduces a selection criterion. As a criterion of selection, it is possible, for example, to have thematic domains. These are hotels, restaurants, curiosities and muses,

  theaters, amusement parks, swimming pools, administrations or companies.

  Other selection criteria may be a region, for example a district, a city, a municipality or a country. It is also possible to ask whether a place chosen as a destination is important, of medium or small importance; these denominations are made according to limits of number of inhabitants for example "small" corresponds to a range of inhabitants between 0 and 50.00O, "average" between

  a range of 40,000 to 100,000 inhabitants, and "large" is a larger beach

greater than 100,000 inhabitants.

  As another criterion, it is possible to use the postal code or the beginning of the postal code, for example the first two digits. The entry of the selection criterion can be done in the form of voice input; in that case

  phonetic data identifying the criterion of selection.

  are recorded for all selection criteria. According to another exemplary embodiment, a selection criterion is introduced by the unit of

o command 5.

  In a next step of entry, one introduces by the middle

  crophone 4 the word as a complete word. Then we have a second step

  in the selection unit 23 during which the computing unit determines the

  These phonetics for which the selection criterion introduced corresponds to complementary information recorded in addition to the phonetic data. According to a preferred example, an index of a data list is established in the working memory 14 of the navigation device 1 in that the respective phonetic data of the data carrier 8 are copied. It is also possible to have several selection criteria

  o not just one, which simplifies the search.

  In a next comparison step 24, the calibration unit

  7 of the navigation device 1 compares the voice input signal made by the microphone 4 for the word to the phonetic data in memory in the working memory 14. In another embodiment, the phonetic data are classified according to the models. voice, so that in addition

  only the phonetic data recorded for the compa-

  reason, and which are close to the introduced vocal model.

  In a next decision step, it is checked whether there is sufficient agreement with any of the phonetic data.

  and the concordance can be influenced by a level of concordance pre-

  determined. When there is no constant concordance, it returns to the entry stage 22 and the user repeats the introduction of the word. This is indicated, for example, by the display unit 3 or by the loudspeaker 2. If it is found that there is no clear association possible but that however 3s several possible phonetic data have a concordance sufficient, we pass on a second selection step 26 by which the user can select an input on the display unit 3 using the keys 6. From the second selection step 26, we arrive at the first definition step 27 during which we set the entry for

following the treatment.

  In a navigation device for introducing a desti nation, the first word to be introduced is generally a city name which is followed by other entries such as a street name and, where appropriate, a building number. In general, the introduction of a

  word is not enough to determine a destination precisely.

  Thus, for example, to enter a hotel or restaurant you are looking for, you first have to give the name of the city and then enter the name of the restaurant. The first determination step 27 can thus be followed by another entry of a selection criterion not shown in FIG. 2. If an address is selected, such an entry is not necessary because, as city is given, we already have a criterion of selection, so that the next entry of a street name will have to take into account only the street names of this city. Advantageously, for this street name, we record the city where this street is located. Since the same street names can exist in several cities, you can only register a wheel name once for different cities with information

  additional corresponding cities.

  The first determination step 27 continues for the entry of a city name by a second entry step 28 during which a street name is introduced as a complete word. In a second decision step 29, it is checked whether it is possible to compare to identify the audio signal introduced and the phonetic data transmitted from the working memory 14. If this is not the case, it returns to the second input step 28. If after a comparison, several recorded phonetic data may be involved, a third selection step 30 is passed through which the user selects several street names. After a selection of one of these street names, we pass on a second determination step 31 during which we fix a entry of a street name for further processing. We arrive in the second determination step 31 from the decision step 29 if during this it was found that it was possible to have a one-to-one attribution between the phonetic data and the word voice signal. intro 3s duit. After the second ascertainment step 31, a processing step 32 is passed during which the information entered is processed, in particular to determine a path between the position of the vehicle.

  obtained by the first radio interface 18 and the introduced destination.

  FIG. 3 shows an example of data recording on the data medium 8. The data are preferably recorded in a data field, a data line 40 being associated with a phonetic data set 41 which is registered for the preterence of the data. for example, in the form of an audio signal in the first cell of the data line 40. A location or name designation 42 is associated with the phonetic data set 41. The name designation 42 may be a city for example "Stuttgart" and for a street for example "Wernerstrasse", and for a hotel, restaurant, administration or other facilities or for example for a company designation for example "Bosch". In the other data fields 43 ',

  43 ", 43" ', the additional information is recorded. The information

  For example, for city phonetic data, a postal code, a district, a country or a region. For the

  the only additional information is the cities in the

  what are the streets. In addition, in a field you can also

  record how many times the phonetic expression has already been introduced.

  This allows a search directed to destinations already frequently introduced, for example all the destinations that were used until then more than five times. It is also possible to record the instant of the last call as additional information. Thus, we first search among the entrances that have already been made during a period

  for example in the last two weeks.

Claims (7)

R E V E N D I C A T IO N S   1) A method of vocal introduction of a word according to which the vocal signal is inputted by a microphone, characterized in that phonetic data with at least one complementary piece of information is recorded, the user introduces a selection criterion, compares the selection criterion introduced with the additional information in memory, the phonetic data is compared to the voice signal for which the selection criterion corresponds to the additional information recorded.   and from the comparison, the introduced word is defined.   2) Method according to claim 1, characterized in that in addition to the phonetic data, location indications are stored and / or name designations.   3) Process according to claim 1, characterized in that   the selection criterion is introduced by a voice input.   4) Method according to claim 1, characterized in that as complementary information, indicates a membership in a   thematic area, to a region and / or a place.    Method according to Claim 1, characterized in that, as complementary information, at least one part is memorized a postal code of a place.   6) Process according to claim 1, characterized in that   as additional information, the frequency of the se- lection of a word made until then.   7) Method according to claim 1, characterized in that the phonetic data are recorded with an indication of place and / or   a name name on a rewritable data carrier.   8) Method according to claim 1, characterized in that phonetic data are recorded on a data medium for the words introduced by voice introduction into the navigation device.   and the recorded phonetic data and the voice signal are compared.   troduct, the phonetic data that can be acoustically